arXiv subjects

Li Zhang

Publications and source records attributed to Li Zhang.

At least 127 records · Page 7Linked to original sources

arXiv2026

RoboRouter: Training-Free Policy Routing for Robotic Manipulation

Research on robotic manipulation has developed a diverse set of policy paradigms, including vision-language-action (VLA) models, vision-action (VA) policies, and code-based compositional approaches. Concrete policies typically attain high success rates on specific task distributions, but limited generalization beyond it. Rather than proposing another monolithic policy, we propose to leverage the complementary strengths of existing approaches through intelligent policy routing. We introduce RoboRouter, a training-free framework that maintains a pool of heterogeneous policies and learns to select the best-performing policy for each task through accumulated execution experience. Given a new task, RoboRouter constructs a semantic task representation, retrieves historical records of similar tasks, predicts the optimal policy choice without requiring trial-and-error, and incorporates structured feedback to refine subsequent routing decisions. Integrating a new policy into the system requires only a lightweight evaluation and does not incur training overhead. Across simulation benchmark and real-world evaluations, RoboRouter consistently outperforms individual policies, improving the average success rate by more than 3% in simulation and 13% in real-world settings, while preserving execution efficiency. Our results demonstrate that intelligent routing across heterogeneous, off-the-shelf policies provides a practical and scalable pathway toward building more capable robotic systems.

cs.RO↗

arXiv2026

Reconfigurable all-optical inference via tunable second-harmonic generation and spin-orbit coupling cascade

Spin-orbit coupling (SOC) is widely exploited as a fundamental mechanism for generating orbital angular momentum (OAM); however, conventional approaches typically lack flexibility and tunability. Here, we introduce a continuously tunable second-harmonic generation (SHG)-SOC cascade mechanism modulated by a spatially movable nonlinear crystal. Under linearly polarized excitation, the SHG-SOC cascade engages synchronously with both degenerate and nondegenerate SHG processes, thereby expanding the OAM spectrum and significantly enhancing the information density and feature-mapping capacity of the optical field. Moreover, the OAM spectral distribution can be continuously reconfigured simply by translating the nonlinear crystal. This deterministic physical evolution, which maps simple OAM modes onto a tunable high-dimensional OAM space, is mathematically analogous to the high-dimensional feature expansion performed by a kernel function of a support vector machine (SVM) in machine learning. Such dimensional expansion can project linearly inseparable input data into a high-dimensional space where they become linearly separable. Exploiting this physics-algorithm analogy, we develop a reconfigurable all-optical inference platform. As a proof of concept, we successfully perform classification tasks, including the recognition of Iris flowers and Palmer penguins. This work establishes a scalable, physically reconfigurable architecture for high-dimensional all-optical computing and neuromorphic photonics.

physics.optics↗

arXiv2026

Flow6D: Discrete-to-Continuous Flow Matching for Efficient and Accurate Category-Level 6D Pose Estimation

6D pose estimation is a key task in computer vision and embodied AI, widely used in robotic manipulation, augmented reality, etc. Existing methods directly regress in a high-dimensional continuous space, facing two key challenges in category-level pose estimation: limited accuracy due to noise and local optima, and inefficient search over an infinite space that hinders real-time performance. This paper proposes Flow6D, a hierarchical flow matching framework with a two-stage discrete latent space localization-continuous pose regression strategy. Rotation and translation parameters are first discretized into bins, with a discrete flow matching model locking the latent space around the true pose to reduce search complexity. Then, by sampling in the latent space, a continuous flow matching model predicts local pose residuals to optimize the estimate and regress to an accurate pose. The framework also naturally extends to articulated objects, outperforming state-of-the-art methods on synthetic and real datasets with real-time inference at 70 FPS. Project website: https://flow6d.github.io/.

cs.CV↗

arXiv2026

Multi-Modal Hyper-Graph Fusion for Low-Light Crowd Counting

Crowd counting is a fundamental task in computer vision. However, crowd counting in low-light environments remains largely underexplored, despite its practical importance in the real world. Existing methods mainly focus on well-lit scenes or rely on single-modality Red-Green-Blue (RGB) representations, which often become unreliable under extreme darkness and complex non-uniform illumination. To handle this problem, we construct three new low-light crowd counting benchmarks, which consist of two synthetic datasets, SHA\_Dark and SHB\_Dark, and a real-world benchmark LC-Crowd (Low-light Crowd Dataset). Inspired by Retinex-based physical modeling, we introduce depth and Canny edge cues as complementary geometric and structural priors to enhance the intrinsic reflectance representation under low-light conditions. We propose a Multi-Modal Hyper-Graph Fusion module, which formulates RGB appearance, depth geometry, and edge structure cues as nodes in a unified hyper-graph and explicitly captures their high-order complementary relationships via dynamic hyperedge construction and message passing. Furthermore, to adaptively allocate computation in dense prediction, we propose a Deformable Rectangular Sparse Attention (DRSA) module, which concentrates computation on informative regions through anchor-aware estimation and adaptive rectangular window modeling. Based on these designs, we develop a unified Low-Light Counting Network (LCNet) for robust low-light crowd counting. Extensive experiments on three benchmarks demonstrate that the proposed method achieves the best overall performance against existing state-of-the-art (SOTA) methods. The code is in the supplementary material. The datasets will be made public upon acceptance.

cs.CV↗

arXiv2026

RT-Counter: Real-Time Text-Guided Open-Vocabulary Object Counting

Text-guided open-vocabulary object counting (TOOC) aims to count objects belonging to the categories specified by natural language descriptions. Although vision-language pre-trained models have been successful applied to TOOC tasks, they still struggle with fine-grained spatial understanding and real-time inference requirements in counting scenarios. To address these limitations, this paper proposes a real-time TOOC framework, called the Real-Time Counter (RT-Counter), that achieves not only good counting accuracy but also high computational efficiency. RT-Counter designs a novel Visual Prototype Textualization (VPT) module that can project learned visual features into a text feature space and then generate features containing the abstract information that is hard to capture with visual prototypes and the detailed prototype information that is difficult to describe in text, enhancing the object-level visual-language model's counting capabilities. Additionally, RT-Counter incorporates our Weaving Transformer (Weaformer) layers, maintaining high descriptive power at a fraction of the computational cost. The Weaformer layer adopts a novel hybrid attention mechanism that can efficiently weave together local and global visual features. Extensive experiments on three public datasets show that RT-Counter successfully breaks the accuracy-speed trade-off in TOOC. While achieving a competitive MAE of 13.30 on FSC147, RT-Counter operates at 112.48 FPS, making it 7.4x faster and over 4$\times$ more parameter-efficient than the existing leading methods in TOOC. Our work aims at balancing high accuracy and real-time performance in TOOC. Code is available at: https://github.com/Jason-Mar1/RT-Counter.

cs.CV↗

arXiv2026

Test-Time Training for Robust Text-Guided Open-Vocabulary Object Counting

Text-guided Open-vocabulary Object Counting (TOOC) enables counting arbitrary object categories specified by text prompts, offering substantially greater flexibility than conventional closed-set counting. However, existing TOOC methods are developed and evaluated primarily on ideal images, while real-world scenes often suffer from adverse conditions such as rain, fog, darkness, and sensor noise, which severely degrade visual quality and impair vision-language alignment. To bridge this gap, we introduce Robust-TOOC, the first benchmark for evaluating TOOC under diverse corruption conditions, which covers six representative degradation types: rain, fog, darkness, Gaussian noise, salt-and-pepper noise, and mixed corruption. To improve robustness while preserving the original counting architecture, we propose Dual-TTT, a dual-architecture test-time training framework for TOOC. Specifically, during test-time training, Dual-TTT updates only the Text-guided Lightweight Denoising module (TL-Denoiser), while keeping the original counting network frozen. Inspired by diffusion models, the TL-Denoiser is optimized to remove corruption-aware noise from image representations under degraded conditions. Since only the TL-Denoiser is trained at test time, Dual-TTT is annotation-free and can be seamlessly integrated into existing TOOC models without modifying their original architecture. Extensive experiments on multiple recent TOOC baselines demonstrate the effectiveness of our method.

cs.CV↗

arXiv2026

MambaCount: Efficient Text-guided Open-vocabulary Object Counting with Spatial Sparse State Space Duality Block

Text-guided Open-vocabulary Object Counting (TOOC) aims to estimate the number of objects described by text prompts, which is particularly challenging in dense scenes with large scale variations. Existing TOOC approaches predominantly rely on Transformers, whose quadratic complexity with respect to image resolution limits their scalability. Mamba offers a promising alternative due to its linear complexity. However, previous Mamba-based methods have two main limitations. On the one hand, the inherent causal formulation of Mamba constrains the bidirectional spatial dependency modeling required by non-causal vision tasks. On the other hand, existing Mamba-based vision models often overlook the unconstrained high entropy in the spatial token responses, which can weaken local details and high-frequency cues. To address these limitations, we propose MambaCount, an efficient framework built on the Spatial Sparse State Space Duality (S^4D) block. Specifically, we analyze and reconstruct the decay dynamics of hidden states in Mamba to alleviate the dependency constraints introduced by causal modeling. Moreover, we introduce a Spatial Token Selection (STS) sub-block to reduce the unconstrained high entropy in spatial token responses within Mamba. In addition, we design Multi-Granularity Prototypes (MGP) to identify object-like regions at different semantic levels, improving cross-modal alignment and interpretability. Extensive experiments on FSC-147 demonstrate that MambaCount achieves state-of-the-art performance among methods without secondary querying, obtaining a test MAE of 12.23, while retaining linear complexity.

cs.CV↗

arXiv2026

A comparative and critical study of EEGNet for fNIRS-driven cognitive load classification

Accurately classifying cognitive load from functional near-infrared spectroscopy (fNIRS) signals remains a significant challenge due to temporal variability, inter-subject differences, and sensitivity to preprocessing choices. This study provides a comprehensive evaluation of EEGNet for fNIRS-based cognitive load classification by systematically examining the effects of temporal segmentation strategies (overlapping vs. non-overlapping), window lengths (10s, 20s, 30s), feature extraction methods (Analysis of Variance (ANOVA), Principal Component Analysis (PCA), Fast Independent Component Analysis (FastICA)), learning rate configurations (fixed and adaptive), and evaluation protocols (random split vs. subject-independent (SI)). Results from random-split experiments show that overlapping segmentation, combined with smaller fixed learning rates (0.01-0.001), yields the highest accuracies, due to temporal redundancy and dense sampling of hemodynamic transitions. However, SI evaluation reveals a substantial drop in accuracy, demonstrating limited generalization to unseen participants. Under SI evaluation, non-overlapping segmentation outperformed overlapping windows, with the best accuracy of 56.11% achieved using PCA features with a 20-second window and a 0.1 learning rate. These findings indicate that eliminating temporal redundancy helps the model learn more robust and generalizable representations of cognitive load across individuals. Although adaptive learning rate strategy improved training stability, it did not surpass the performance of optimally selected fixed learning rates. The study highlights the critical role of segmentation strategy and learning rate selection in improving model generalization and identifies methodological considerations essential for developing reliable, real-time, and SI cognitive load classification systems using fNIRS.

cs.LG↗

arXiv2026

3D Consistency Optimization for Self-Supervised Monocular Video Depth Estimation

Reliable monocular video depth estimation is crucial for downstream 3D reasoning and embodied AI in endoscopic navigation. However, existing self-supervised approaches typically treat video frames independently or rely on weak temporal regularization. These methods, lacking a holistic perception of the underlying 3D scene, inevitably suffer from geometrically inconsistent predictions and severe cross-frame drift. To address these limitations, we introduce a new paradigm that recasts sequential video depth estimation as an unconstrained multi-view 3D reconstruction problem, enabling full exploitation of the powerful geometric priors embedded in recent 3D foundation models. The core of our approach is a 3D consistency optimization framework driven by three constraints: image-level photometric rendering, explicit world-coordinate geometric alignment, and multi-scale temporal gradient consistency. Such unified optimization elegantly anchors isolated frames to a globally coherent 3D structure. Our method has been validated in both the self-supervised training scenarios and challenging zero-shot clinical environments. Results show that the proposed approach achieves state-of-the-art spatial accuracy, outperforming the frame-based, video-based depth estimators and the multi-view 3D reconstruction baselines.

cs.CV↗

arXiv2026

Metis: A Generalizable and Efficient World-Action Model for Autonomous Driving and Urban Navigation

World action models~(WAMs) have shown great promise for autonomous driving and urban navigation. Built upon Vision-Language-Action models or video generation models, existing approaches suffer key limitations: (1) High inference latency due to future observation prediction at test time, and (2) tightly coupled video and action modeling leading to representational mismatch and degraded generalization. To address both issues, we propose Metis, an end-to-end WAM framework that decouples video generation and action prediction. Specifically, Metis employs a Mixture-of-Transformers architecture with dedicated experts for video generation and action prediction, preserving the intrinsic distributional properties of each task. To enhance efficiency, we introduce an asymmetric attention mask that enables joint training of both experts while allowing the action model to bypass explicit video generation during inference. This design ensures training-inference consistency and significantly reduces computational costs without compromising planning performance. Extensive experiments demonstrate state-of-the-art performance on the NAVSIM navhard and navtest benchmarks and the CityWalker navigation benchmark, validating both the generalizability and efficiency across diverse tasks. Real-robot deployments further confirm the practical feasibility of our approach.

cs.CV↗

arXiv2026

Self-Evolving Vision-Language Models for Image Quality Assessment via Voting and Ranking

Improving vision-language models (VLMs) in the post-training stage typically relies on supervised fine-tuning or reinforcement learning, methods that necessitate costly, human-annotated data. While self-supervised techniques have proven effective for enhancing reasoning capabilities, their application to perceptual domains such as image quality assessment (IQA) remains largely unexplored. In this work, we introduce EvoQuality, a novel framework that enables a VLM to autonomously refine its quality perception capabilities without any ground-truth labels. EvoQuality adapts the principle of self-consistency to the ranking-based nature of IQA. It generates pseudo-labels by performing pairwise majority voting on the VLM's own outputs to establish a consensus on relative quality. These pseudo-rankings are then formulated into a fidelity reward that guides the model's iterative evolution through group relative policy optimization (GRPO). By iteratively leveraging its own predictions, EvoQuality progressively refines the VLM's perceptual capability. Extensive experiments show that EvoQuality boosts the base VLM's zero-shot performance by 31.8% on PLCC across diverse IQA benchmarks. Remarkably, despite being entirely self-supervised, EvoQuality achieves performance that is competitive with, or even surpasses, state-of-the-art supervised VLM-based IQA models, outperforming these models on 5 out of 7 IQA benchmarks. Furthermore, the framework demonstrates significant flexibility, allowing it to be stacked with pre-trained IQA models to bolster generalization on unseen datasets. Codes and checkpoints will be available at https://github.com/bytedance/EvoQuality.

cs.CV↗

arXiv2026

DLawBench: Evaluating LLMs Through Multi-Turn Legal Consultation

Lawyer-client consultation is a critical starting point for legal services. Effective legal assistance hinges on eliciting sufficient and truthful information from clients in order to devise strategies that best protect their interests. This task requires Large Language Models (LLMs) not only to perform robust legal reasoning, but also to strategically elicit material facts through multi-turn interactions and effectively guide clients with diverse personalities. Yet existing legal benchmarks overlook this interactive capability. To fill this gap, we introduce DLawBench, a diagnostic benchmark for real-world legal consultation. Drawing on realistic client behavior, we characterize lawyer-client interactions into four types: Cooperative, Dependent, Withdrawn, and Adversarial. Using dialogues grounded in real cases, DLawBench evaluates whether LLMs can effectively conduct legal consultation under realistic conditions. DLawBench comprises 461 cases from Chinese and U.S. law, 5,532 paired fact entries, 3,411 inquiry rubrics, and 3,348 issue-resolution rubrics, and evaluates 26 representative LLMs. Systematic experiments show substantial headroom: the best-performing model, GPT-5.5, achieves only 0.562 on consultation-grounded legal reasoning. More importantly, DLawBench exposes both sycophancy in legal consultation and a paradox: models perform worse when clients need guidance most.

cs.CL↗

arXiv2026

Observation of localization reversal and harmonic generation in nonlinear non-Hermitian skin effect

The interplay between band topology and material nonlinearity gives rise to a variety of novel phenomena, such as topological solitons and nonlinearity-induced topological phase transitions. However, most previous studies fall within the Hermitian regime, leaving the impact of nonlinearity on non-Hermitian topology seldom explored. Here, we investigate the effects of nonlinearity on the non-Hermitian skin effect, a well-known non-Hermitian phenomenon induced by the point-gap topology unique to non-Hermitian systems. Interestingly, we discover a nonlinearity-induced point-gap topological phase transition accompanied by a reversal of the skin mode localization, which is distinct from previous nonlinearity-induced line-gap topological phases. This phenomenon is experimentally demonstrated in a nonlinear microwave metamaterial, where electromagnetic waves are localized around one end of the sample under a low-intensity pump, whereas at a high-intensity pump, the waves are localized around the other end. Our results open a new route towards nonlinear topological physics in non-Hermitian systems and are promising for reconfigurable topological wave manipulation.

cond-mat.mes-hall↗

arXiv2026

Foveated-Imaging Geometry CT Architecture and Seeded Diffusion Model Enabling Global Super-Resolution Reconstruction

For X-ray computed tomography (CT), a smaller detector pixel size generally leads to higher scanner spatial resolution, but inevitably increases system cost as well as data overhead in acquisition and processing. To achieve high-resolution (HR) CT imaging in a more resource-efficient manner, we propose a Foveated-Imaging Geometry CT (FIGCT) architecture, which integrates local HR data into an acquisition scheme dominated by low-resolution (LR) measurements. We further develop a Diffusion Probabilistic FIGCT Super-Resolution Reconstruction (DPFSR) framework to generate global HR CT images over the full field of view (FOV). The concept of FIGCT is first established, and its typical configurations are characterized according to the arrangement of HR data. Two key indices, namely the HR data fraction (HDF) and the LR-to-HR detector pixel size ratio (LHR), are introduced to describe the FIGCT geometry. The proposed DPFSR incorporates local HR information into intermediate clean-image estimates in both the projection and image domains during the reverse diffusion process. This additional step not only guides HR image generation from LR data, but also improves data consistency between the clean-image estimates and the originally measured data. Preliminary numerical simulation results on FIGCT show that the proposed architecture provides high-precision CT images within the region of interest (ROI) corresponding to the HR data, while the spatial resolution deteriorates rapidly outside the ROI. With DPFSR, global HR reconstruction is achieved on the AAPM Grand Challenge dataset and swine lung CT data, outperforming existing SR methods in terms of Learned Perceptual Image Patch Similarity (LPIPS), PSNR, and SSIM.

physics.med-ph↗

arXiv2026

Gamma-Ray Emission from the Crab Pulsar: A 17-Year Fermi-LAT Reanalysis

We present a reanalysis of 17 years of Fermi Large Area Telescope (LAT) observations of the Crab pulsar obtained between 2008 August and 2025 August. Using monthly Jodrell Bank radio ephemerides, we assigned pulse phases to the LAT events and aligned the phase zero across the full data set. From this phase-aligned data set, we derived pulse profiles over 100 MeV to 300 GeV. The pulsed emission remains clearly detectable in the 10 to 20 GeV and 20 to 30 GeV bands, with H-test significances of 32.36 sigma and 11.59 sigma, respectively, but is not significantly detected in the 30 to 300 GeV band. Phase-resolved likelihood analysis was performed over 100 MeV to 30 GeV using 14 phase bins with comparable pulsed statistics. The fixed-window fractional fluxes show that the contribution of Peak 1 (P1) decreases steadily with energy, while those of Peak 2 (P2) and the Bridge increase, with P2 exceeding P1 above 10 GeV. Finally, the same phase-assignment framework also enables an off-pulse analysis from 100 MeV to 1 TeV, confirming the synchrotron and inverse-Compton components that dominate the emission in the selected off-pulse interval.

astro-ph.HE↗

arXiv2026

MADE: Beyond Scoring via a Multilingual Agentic Diagnosing Engine for Fine-Grained Evaluation Insights

Multilingual and multicultural benchmarks now cover dozens of languages and model families, but the resulting score landscapes remain metric-rich and insight-poor, necessitating fine-grained multilingual post-evaluation diagnosis. However, single LLMs and open-ended agents are easily swamped by the long, noisy diagnostic input, and no reusable taxonomy exists for it. To address this, we propose MADE, a Multilingual Agentic Diagnosing Engine that decomposes post-evaluation analysis into planning, aggregate analysis, instance-level case inspection, multilingual and cultural reflection, and grounded report synthesis. MADE is paired with an expert-led 54-query and 15-language diagnostic set, evaluated on top of a large-scale multilingual evaluation substrate (33 model families, 11 benchmarks, 26 languages, 34 cultures, 8.66M evaluation records). Experiments show that MADE outperforms the strongest shared baseline by 47% in diagnosis report quality and is preferred by human multilingual experts in 87.9% of pairwise comparisons. Applied with multilingual experts, MADE further surfaces four actionable findings on deployment, iteration, and cross-cultural pitfalls, turning benchmark score tables into model-selection and remediation guidance.

cs.CL↗

arXiv2026

EEGDancer: Dynamic Emotion Latent Space Masked Modeling with Reinforcement Learning for EEG Continuous Emotion Prediction

Continuous electroencephalography (EEG) emotion prediction aims to model the temporal evolution of human emotional states from EEG signals. Unlike conventional discrete emotion recognition, continuous prediction requires capturing long-range temporal dependencies and coherent emotional dynamics. However, existing methods mainly rely on point-wise regression and directly model noisy high-dimensional EEG features, limiting their ability to characterize continuous emotional evolution.To address these challenges, we propose EEGDancer, a dynamic emotional latent space learning framework for continuous EEG emotion prediction. The framework integrates vector-quantized representation learning, masked temporal modeling, and reinforcement learning-based trajectory optimization into a unified architecture.Specifically, a causal spatiotemporal Vector-Quantization Variational Autoencoder (VQ-VAE) is designed to learn structured emotional prototypes and construct a discrete-continuous emotional latent space from EEG signals. Based on the learned latent representations, a Transformer-based masked dynamic modeling strategy captures long-range emotional dependencies and temporal evolution patterns. Furthermore, continuous emotion prediction is formulated as a sequential decision-making problem, and a Soft Actor-Critic (SAC) framework is introduced to optimize emotional prediction trajectories at the sequence level instead of frame-wise local fitting.Extensive experiments on the SEED, SEED-IV, and Long-Term Naturalistic Emotion datasets demonstrate that EEGDancer consistently outperforms existing machine learning and deep learning methods. Ablation studies further verify the effectiveness of the proposed latent space and reinforcement learning-based trajectory optimization for modeling continuous EEG emotional dynamics.

cs.HC↗

arXiv2026

Modeling Gamma-Ray Burst Spectra with Convolutional Neural Networks: Fast-Cooling Synchrotron Emission in a Decaying Magnetic Field

The radiation mechanism of gamma-ray burst (GRB) prompt emission remains uncertain. Although the fast-cooling synchrotron model in a decaying magnetic field can account for the characteristic nonthermal spectral shape, its computational cost has limited its use in systematic observational fitting and statistical model comparison. We develop a convolutional neural network (CNN)-based spectral emulator for this physical model and train it on a large synthetic data set generated over a physically motivated parameter space. The trained network reproduces the numerical spectra with high fidelity while reducing the cost of spectral evaluation to the millisecond level. We then incorporate the emulator into a Bayesian spectral-analysis framework and apply it to the time-resolved spectra of GRB 231020A observed by Fermi/GBM. In most time intervals, the decaying-field fast-cooling synchrotron model provides better fits and smaller Bayesian information criterion values than the standard fast-cooling synchrotron model. These results suggest that a radially decaying magnetic field provides a plausible and more physically motivated interpretation of the prompt-emission spectrum of this burst, while also indicating that the emulator offers a practical route for large-sample Bayesian inference and systematic comparisons of GRB prompt-emission models.

astro-ph.HE↗